A media selection calculus
John D. C. Little, Leonard M. Lodish · DSpace@MIT (Massachusetts Institute of Technology) · 2011
Computing costs have been a fraction of a percent of the cost of the media scheduled.Improvements over previous schedules, as calculated by the model from the user's input data, have run from 5% to over 20%.It is because of the multiplicity of seemingly reasonable choices usually available, because of the complexity of advertising phenomena, and because of the quantity of media decisions that are made.Our goal is to build a media model that will increase advertising productivity.This requires that the model lead people to make better media decisions; it requires the model to be economical to use; and it requires that the model be, in fact, used.To establish that the model will increase productivity is, to say the least, difficult.Certain of the required inputs will be subjective.Many aspects of the advertising effectiveness process are poorly understood.The most satisfactory test of validity would be to predict outcomes (e.g., sales) and compare them with actual results, but the inherent variability in sales and the problem of relating sales to advertising when other marketing variables and competition also affect response make this difficult.Considering these obstacles, perhaps we should give up, at least until the underlying processes are better understood.Media planners obviously do not have this option.They must do something sensible with the informa- tion they have.Furthermore, they have to do this in the midst of day to day pressures.The important questions then are: Can we isolate the most relevant phenomena for media planning, can we put them together into a consistent structure, and can we link the media planner to the structure in a practical way that increases his power to think about the problem?We shall argue that the answers to these questions are yes.What then are some of the facts and phenomena relevant to media selection?We have said that the main purpose of media is to deliver messages to potential customers efficiently.Relevant to this are at least the following ideas :